October 1949
2221
I N D U S T R I A L A N D E N G I N E E R I N G C H E.M I S T R Y
QYKTHESIR O F ISOPROPYLBESZESE. By the ester iiiethod iS0proI)yliuethniiesulfonate (1.06 moles), berizeiie (3.6 moles), and metlianesulfonic acid (0.5 mole) were heated to 80" C. After an sudden reaction occurred, and induction period of 10 minut the mixture sepnrated abrupt to txvo phases. The tenipera11 additional 20 miiiutes, after ture was held a t 80-100" C.
By tlie hot acid method mixed '3770 :ill-licstraigh t-chain acids from acetone, water, and Shell solvent is reported. The adsorption increases in the order given. The bearing of these results on various theories of the adsorption of acids
by resinous anion exchange materials is indicated. Some qualitative and semiquanti tative results for the adsorption of a variety of acids from hydrocarbon solution on the same adsorbent are given, with a few suggestions f o r practical application and reports of such applications.
IiY
adsorption of fatty acids from nonaqueous solvents. h i t i i k study attention is directed t o the straight-chain carbosylic acids. For comparative purposes the adsorption of carhos!-lic acids froin aqueous media also is shown. The adsorption of these acids froin iioii:iqueous media may be either grcs:tter or smaller thnii t h e :idsni~pt ion of the eo1 pontling acids from 1r:Lter.
T-IETT- oi tlie 1-arious opinions ( 2 , .$, 7 ) as to the mechanism of acid adsorption on anion exchange rei.ins. the invcstig:itioii of these adPorbi'nts was erteritied to solvc~ntsother thnn n-ater. Such :in invehtig:ition also mi?-open :idtlition:il fields of npplication for thew interesting adsorl)ents. Enrlier ~ o r kit1 this field by XJ-ers (6) usiiig Ahnl)erliteIR-4 anion eschmge resin indicated th:it some water TI-as rrquii,ed for effective adaorption of the fatty acids from solvents such :is benzenc. .'Iniherlitc IR-lB dries t o hard gl:i~sysolid which appnreritl?- is lt~rgelyimpervious t o solvents such as benzene a n d thus presents little active surface for acid adsorption when conipletcly d q . Duolitc -1-2; on the other hand, remains relatively porous 011 drJ-ing and it has long heen knoivn th:it with this resin no n-ater is necessary for the
EXPEKI3IESTA L
~IITCKIIL Tlw ~ . .rc9sin choscin for this study wits Iholitc. -1-2. manufactured by tlie C'licmicnl Process Coinp:iny. Tho Duolitc A resins diffrr from t other coinmcreid anion exchange resins in that they remain ively porous on drj-ing and thus presmt a much larger active, sui,face than do rwins ivhich shrink grcntly i n
INDUSTRIAL AND ENGINEERING CHEMISTRY
2222
Vol. 41, No. 10
solvent experiments, the titration method used was t h a t recommended by the rZmerican Society for Testing Materials ( 1 1. RESULTS A S D DISCUSSION
Figlire 1.
The results have been condensed into four graphs and two tables. I n Figures 1 to 3 the results were plotted according t o t,he Freundlich adsorption equation, x / m = KC"". The points of the log-log curve actually plotted were taken from the smoothed curves obtained in plotting x / m against c, where x and m are grams of adsorbed solute and adsorbent, respectively, and c is given in moles solute per liter of solution. T h e actual esperimtwtal points fitted the foregoing curves closely except a t t h r very lowest concentration where the titration errors could be expectrd to he great. This may be seen in Table I where the actual esperimental results are given for the adsorption of propionic acid from Shell solvent. These results are entirely typical of all the other results shown in Figures 1 to 3. I n Figure 4,the e p / m (obtained from graphs of x/mM = k'C1/n, where 31 is the molecular weight) adsorbed a t C = 0.25 is plotted against the number of carbons in the sorbed acid. The adsorption from Shell solvent and acetone decreases with increase of length of carbon chain (reversal of Traube's rule). I n water the adsorption appears to be substantially constant except for formic acid, which is a much stronger acid than the remaining members of the series and would be expected t o form less easily hydrolyzable salts with the amine groups in the resin.
Sorption of Acids from Acetone
drying. The resin (sieved to -20+35 mesh) use by first wetting the commercial product with n-atcr and thcri cvcling twice with excess aqueous hydrogen chloride ( 1 0 5 strength) and sodium hydroxide (5 to 10%). The excess alkali was Ivashed out x i t h demineralized water and the resin suctionfiltered. It was then dried, first at room temperature and trlallg in an oven a t 50" C. for 24 hours. Test shoned that the resin dried to constant weight within this pt'riod. Although this does not prove that such treatment completely frees the resin from the last tracer of water, resin dried a t 100" C. show3 substantially the same acid sorbing properties from nonaqueous solveiits. The latter temperature was not employed as the resin on drying a t this temperature will eventually show a sniall rcgain in wcight indicative of mild oxidative changes in the resin. The dry resin was filled into well dosed bottles which were only opened as required for use. Midway in these experimcnts it becitine necessary t o prepare more dry resin. The second batch was treated as the first and check runs were made to see horn it compared to the first batch. These expcrimcnts (not otherwise reportrd) shon-ed suhstantial agreement between the separate batchrh of resin. The straight-chain carboxylic acids used w r e all Eastmail I:.P. grid? with the exception of the lauric acid which was a Kahlbaum product of similar grade. The acids w r c riot Eurthcr purifird. The Shell solvent (a Stoddard type solveiit) uscd was a co1xiniereial product. As received it had a slight ycllon- color but this was readily removed by treatment with activated carbon. T h r solvent was dried by passage through a colunin of Drierite (:inhydrous calcium sulfate). T o eliniiriate traces of moisture in the avid solutioiis made up with this solvent, th(: stock solutions of acid in the solvent m r e also passed over Drierite before annly Solutions of various concentrations were then preparcd by dilution of tlie stock solution with appropriate aniounts of dry solvriit. The itcetoiie used was C . P . grade aiid was not further purified. It was dried over calcium chloride. The water uscd tvxb ordinary once-distilled water from the regular Laboratory 5upply. A ~ E T R O D O F STLUY. I n all esperimeiits, the weighed amount of resin was about 15.0 grams. Thc resin was transferred to a clean dry bottle and the required amounts of solvent and solution added to give a tot,al volume 01' solution of 200 ml. The bottles were immediately stoppered and placed i n a constant temperature bath :it 26" C. and gently agitated for n period of 21 hours. i n separate espeiiments this period of time n u shown to he more than adequate for substantially complete equilibrution. The equilibrated samples were removed from tlie bath and 10-ml. sliquots taken for analj
TABLE
I.
S O R P T I O S O F P R O P I O S I C X C I D FROM S H E L L SOI.\~EN'I B Y DUOLITE A-2
V-eirlit of Ko.in, G . 15.904 14.858
14.588 15,992
13 428 14.3!17 14,!382
Equivalents Sorbed 0,0608 0 . 08.70 0.1012 0.1203 0,1229 0.1251 0,1288
x/m
(G./G.) 0.348 0.424 0.516 0.559 0.590 0,636 0.642
e (Moles Sn1,utr I.iter Solution) 0.00909 0.0453 0.1210 0.1827 0,3260 0.4720 0.6060
Figures 1 to 3 show that the results fit the Freundlich equatiori quite xell except a t the higher concentration where there is a definite tendency toward a decreasing slope. As a matter of fact. although the process is described as adsorption and the results are plotted according to the Freundlich adsorption isotlierm, thi3 is merely a matter of convenience of presentation; the principal mechanism of binding of the acids is a. chemical interaction with the formation of a covalent bond betn-een the amine The curves obtained groups of the resin and the acid bond. simulate a true adsorption process largely because the amines in the resin present a wide spectrum of basicities (3). The resins may therefore be expected to show a smoothly increasing binding of the acid with increasing concentration of the acid up to some
-.2 --
0
-.4
0 -I
-.6-
-.O -
,
INDUSTRIAL AND ENGINEERING CHEMISTRY
October 1949
2223
tone, whereas in water the resin swells about 17%. Accordingly, the amine groups in the water-wet resin would be expected t o be substantially more available for acid adsorption initially. I n 0 order to overcome the initial handicap with respect to water, the resin must be swollen by the adsorption of acids when working -. 2 with acetone and Shell solvent. This clearly requires a greater driving force of reaction for the latter cases. With Shell solvent, -.4 it appears that the AP' is more than sufficient to swell the resin, XlZ rvhereas with acetone the AF is not so large and swelling is relaW 0 tively small. The difference in AF for the two solvents is, of -I -.6 course, related to the competition between the resin and the solvent for the solute acid and it is to be expected that acetone, the better solvent for these acids, would have less tendency to release -.8 the acidq to a competitor. I -14 -12 -IO -8 -6 -4 -2 0 The I n t figures in Table I1 relate to the volume of the resin after ,OG c first wetting with water and then replacing the water with apk i g u r e 3. Sorption c p f kcidc Cram *hell Solvent proximately 10 volumes of first acetone and then Shell solvent: the resin tends to retain its swollen volume. It would have been saturation value, determined either by the total amine content of interesting to determine the effect of this preswelling on the adthe resin or steric considerations, rather thnn the sharply disconsorption of acids from acetone but the work was completed before tinnous type of curve characteristic of the intrraction of two pure it was realized that this matter might be of some importance. cheniiLaal entities. Another factor favoring the type of curve obADSORPTION OF OTHER ACIDS tairird is the sn-clling of the resin on acid adsorption. This could be expected to expose additional amine groups for effective acid I n addition to t,he foregoing studies of the carboxylic acids, some adsorption by rrducing the steric hintlrnnce to such binding of the additional qualitative and semiqusntitative work \vas done on the acid. adsorption of a diverse arral' of acidic materials from hydrocarbon From the results shown with the Diiulite -1-2 type amine resins, type solvents. I n general, these acids are more strongly adsorbed the adsorption of carboxylic acids from hydrocarbon type solvent from hydrocarbon solvents than from water. For esample, pheis greater than from water, whereas with acetone the reverse is nol, resorcinol, catechol, hydroquinone, and tert-butyl catechol are the case. Inasmuch as the two nonaqueous adsorptions took all adsorbed strongly by the resin, although tert-butyl catechol place in the complete absence of water, it is additionally clear that appcars to 11:ive n lower affinity for the resin than the others. anion exchange, per se, is not involved hrre. The mechanism is Some of the above phenolic compounds are effective inhibitors accordingly s u b s t a n t i d y a covalmt binding of n-hole acid molefor polymerizable monomers, and the removal of these substances cules by the amine groups of the resin, es proposed by Bishop ( 2 ) . from the monomers (actual test with styrene) by treatment with Some true physical orption may occur but it must be minor, the w i n has been accomplished. relative to the foregoing mechanism. Guniii and IIyers ( 4 ) disagree rvith Bishop and present much evidence for an anion eschange mechanism in the adsorption of aclids from water. While this is certainly the mechanism when anions such as chloride ion 12c are exchanged from the amine hydrochloride form of the resin for dulfate ion, the argumrnt presented is not convincing as t o the mechunimi of adsorption of acids by thr free amine form of the resin. I t is probable that both nnion exchange and covalent adsorption of whole acid molecules are involved. The latter, from the results here reported, is certainly possible, and it appears to the authors that it is very likely the more important of the two possibilities. I I l l , I / ,
I
4
t
i
'I'XBLE 11. VOLUMEOF 25.O-G~airSAMPLES OF DUOLITE A-2 IN
VARIOUSSOLYESTS
D r y resin Shell .solvent Acetone TVater Wet TT-ith water displaced with acetone Acetone t h e n displaced w i t h Shell solvent
hll. 61.5 61.0 61.0 72.0 71.0 68.0
The greater adsorption of the carboxJ-lic. acids from hydrocarbon type solvents, as compared t o water, is readily explicable. In the latter case the hydrolytic effect of the water on the salts formed between the acids and the very weak amine bases in the resin is a limiting factor on the amount of acicl bound a t any acid conceiitr:ition. In the case of hydrocarbon type solvents this effect is absent. The lorn adsorption from acetone is more difficult to explain. Holyever, with the aid of the data in Table 11, a partial explanation may be offered. In Table I1 the tamped volume obtained by wetting the resin with the foregoing solvents is given. T h e resin does not swell in either Phpll solvent or ace-
2 3 4 5 6 7 8 9 1011 N O CARBON ATOMS IN THE ACID
12
Figure 4. Adsorption of Acids at C = 0.23 11 as a Function of Chain Length of Acid
X semiquantitative test was carried out by passing ii solution of phenol in dry benzene through a column of the dry resin. The resin completely removed the phenol up t o a break-through capacity (first appearance of a trace of phenol in the efluent) of about 0.6 mole pcr liter of resin. I n another experiment the resin with adsorbed phenol was washed with the phenol-free solvent until no phcnol appeared in the effluent. Acetic wid dissolved in dry benzene was then passed through the resin. Phenol immediately appeared in the effluent showing that, the acetic acid was adsorbed more strongly and vias displacing the phenol from the resin. SUGGESTED USES
Further tests have shown that acids such as hydrogen chloride, sulfur dioxide, and aluminum chloride are also adsorbed from hydrocarbon type solution by the dry resin. Such applications have indeed received small scale pilot plant investigation; in
2224
INDUSTRIAL AND ENGINEERING CHEMISTRY
one case tlie resin is being used to remove a fluoride compound, probably hydrogen or boron fluoride, from the products of a reaction catalyzed by such substances. The amine resin aluminum chloride product might be of interest in itself as a catalyst, particularly in view of the report of Kacker ant1 Pines (8)on the eii'ect of water on aluminum chloride catalyzed reactions. The Duolite anion exchange resins are also effective in the renioval from nonaqueous media of whole salts whose cations arc capable of forniing complexes with amines. For example, copper resinate is effectively removcd from solution in yasolirie arid anhydrous ferric chloride from solution in ethyl ether. Sumerous metal ions n-hich form hydrates rather than ammines in water solution may he removed by ammine formation IT-ith the aniinc resin in the absence of nater. I n ammiiic farmatioil the nirtitl cation is, of course, an acid according to thc gc.ncralizctl roiiccpt of acids.
Vol. 41, No. 10
Finally, it may be meritionetl tliat Duolite .I-2 is also an cfTec,tive atlsrbeiit for tlie removal oi acidic or niuiue r e x t i v c s u l ~ t ~ $hme of this n-arkia r c p J r t d i r i :i stances from the g ~ phas('. previous paper ( 5 ) . LITERATLRE CITED
(1) Ana. SOC. Testiiqj .lI&riaZs, Stnntlnrds, part 111, 11. 954 ( 1 9 1 2 ) . ( 2 ) Bishop, J. H., J . Phi/s. Chem., 50, 0 (1946). (3) I'uo;s and Straus, J . Polijrne~.Sei.. 3, 246 (194s). (1) Kuiiin and Myers, J . rlm. Chem. SOC.,69, 2 S i 4 (1948). (5) Mills, G . F., U. 6. Dept. of Conirnerce. OTS, JTn-;hitigtorl, I).
(
R e p t . PB 13608. (6) Llyors, E'. J., IND. Est:. C H E X . . 35, 563 (194a). (7) Schwarte et al., I b i d . , 32, 1462 (1940). (5) Wacker and Pines, J . Am. C'iiem. Soc., 68, 1042 (1946)
SeDaration of Gas-Oil and Wax I Fractions of Petroleum bv Adsorption I J
BEVERIDGE J. _\LAIR, ALBERT J. S\YEET>IAN, . ~ N DFREDERICK D. ROSSINI \-ational Birreuir of Standards, W'ashington 25, D . C .
Results are reported on the separation of the gas-oil and wax fractions of petroleum by adsorption with silica gel. In a single-pass operation, the gas-oil fraction of petroleum can be separated into three portions-namely, a portion which is a mixture of paraffins and cycloparaffins (naphthenes), a portion w-hich is largely mononuclear aromatics, and a portion w-hich is largely polynuclear aromatics. Similarly, the wax fraction (aromatic-free
but containing some oily constituents) can be separated into two portions-namely, a portion w-hich is largely paraffins and a portion which is largely cycloparaffins (naphthenes). Data are also giFen on the effect, in the separation of the gas-oil, of changes in several of the factors affecting the separation. h simple analogy between fractionation by adsorption and fractionation by distillation is drawn.
A
10, 12, 13, and 14, and from 8 to 20 nil. per hour for columns 7 an(! 20. These pressuies, which depended oii the length of the packed section and on the temperature of the operation, varied from 110 pounds per square inch gage for colmiins 1, 2, 3, and 4 t(J 15 poundc per square inch gag? for column 20.
R E C E S T report ( 2 ) from this laboratory described the results obtained on the separation of the kerosene fraction of petroleum by adsorption. A previous report ( 1 ) gave results on the separation of the gasoline fraction. This paper describes the separation of the gas-oil and wax fractions of petroleum by adsorption. Some data are also given on the effect, in the separation of the gas-oil, of changes in several of the factors affecting the separation including the temperature of operation, diameter of the fractionating section, and the length of the fractionating section. A simple analogy between fractionation by adsorption and fractionation by distillation is drawn. METHOD AiXD APPARATUS
The method and apparatus used in this inwstigatioii were essentially the same as those previously described ( 1 , 2 ) . The importaiit characteristics of the adsorption columns, used in the experiments reported in this inveetigation, are given in Table I. The adsorimit n-as fresh silica gcl, rrith a particle size distribution such that 60% v a s betn-een 200- and 325-mesh. TKOdifferent lots of adsorbent were used, one of which gave a slightly better separation than the other. However, in the investigation of any one variable, only one lot of adsorbent wits used. The desorbing liquid used for the gas-oil and --as fractions of petroleum m r e , respectively, n-hexanol and cyclohexanol. The inert gas pressure applied t o the columns was adjusted to produce a rate of flow of liquid varj-ing from 30 to 90 ml. per hour for columns 1, 2, 3, 4,
MATERIALS ISVESTIGATED
The gas-oil and wax fractions of petroleum studied in tliis work were from the Ponca, Olila., petroleum which has been undpr i n vestigation by the .1.P.I. Research Project 6 at this burexu qiiicc. 1028 (8-10). The gas-oil fraction x a s arbitrarily selected to include all the. hydrocarbon material of the original petroleum that ivould 1x5 expected to boil between 230" and 300" C. a t a pressure of 1 atmosphere. Actually, the material was not subjected to tempc,r:itures in escess of 200" C. The fractionation of lower boiling material from the gas-oil had been previously completed in connection Kith the Tvork on the kerosene fraction of the same petroleuni ( 4 , 5 ,11-13). The fractionation of all of the gas-oil, including the heavy end, was performed a t a suitable low pressure by the Gulf Research and Development Company, Pittsburgh, Pa. The wax fraction was prepared from one of the original distillate fractions ( S o . 24) of the lubricant portion of petroleum (3, 8-10) by cstmction n i t h liquid sulfur dioxide a t room tempcrature, by cry~tallizationof the resulting raffinate portion x i t h ethylene chloride a t -18' C., and by filtration of the moltcri was
